A Study on the Number ot Preferable Neighborhood Vectors When Using the Improved Jacobi Matrix Estimation Method for Roll Motion Prediction of Small Boat

摘要

Compared with other vessels, small vessels such as fishing boats undergo significant disturbance even under relatively small wave heights. Particularly, for small fishing boats, if. a large-angle roll occurs while a fisherman is working on the deck ofa parked boat. he may falling into the sea, even if capsizing does not occur. Therefore, understanding the .predictability of the time-series data of ship roll motions in waves is important when investigating seaworthiness. Ueno et al showed that the prediction method of Farmer and Sidorowich is effective for the time-series data of ship roll motion in the waves of the narrow-band response of a resonance system, which is stochastic data，and that it is superior to predictions using the conventional linear autoregressive model. In stochastic data such as the time-series data of the narrow-band response of a resonance system, because the number of neighborhood vectors changes as time progresses, the calculation may become unstaole. Therefore, Ueno and Fan used a method that maintains the number of neighborhood vectors. In this study, we analyzed the number of neighborhood vectors and the prediction accuracy. It was confirmed that the optimum number of neighborhood vectors exists.